The effect of spell-check on the essay quality of primary school ...

<strong>The</strong> <strong>effect</strong> <strong>of</strong> <strong>spell</strong>-<strong>check</strong> on the essay quality <strong>of</strong> primary school children
Kirsty Walter 1 and Dr. Vince Connelly 1
1
Oxford Brookes University, UK
1. Introduction
It is widely accepted that writing fluency and <strong>spell</strong>ing ability
constrain compositional quality in children's essays. Despite the
usefulness <strong>of</strong> word processors for those with learning difficulties,
Connelly, Gee and Walsh (2007) found that primary school
children wrote better essays by hand than by keyboarding.
However, <strong>spell</strong>ing ability was not accounted for and so poor
<strong>spell</strong>ers may have benefitted more than their peers due to the
presence <strong>of</strong> <strong>spell</strong>-<strong>check</strong> in the keyboarded condition (MacArthur et
al, 1996).
<strong>The</strong> current study investigated the <strong>effect</strong> <strong>of</strong> <strong>spell</strong>-<strong>check</strong> on
children's essays. It hypothesised that children's compositional
quality and error rates would be greatest in the keyboarded essays
without <strong>spell</strong>-<strong>check</strong>. Children’s familiarity with <strong>spell</strong>-<strong>check</strong> when
keyboarding may enable them to focus on other higher-level
processes.
A further question is the impact <strong>of</strong> keyboarding fluency on
children’s keyboarded essays.
3. Results
Compositional Quality
Transcription Method
– Grammar & Usage (F(2,64)=4.484, p=0.015)
• Spell-<strong>check</strong> > Handwritten and No Spell-<strong>check</strong>
– Capitalisation & Punctuation (F(2,64)=4.431, p=0.016)
• Spell-<strong>check</strong> & Handwritten > No Spell-<strong>check</strong>
– When controlling for keyboarding fluency and <strong>spell</strong>ing ability –
no <strong>effect</strong> <strong>of</strong> transcription method
– Correlations <strong>of</strong> at least 0.43 between keyboarding fluency and
the CQ for each essay.
Keyboarding Fluency
– Predictive <strong>of</strong> 12-44% <strong>of</strong> variance in most WOLD scores across
transcription methods.
– Did not predict: Grammar & Usage, Capitalisation &
Punctuation for any transcription method, nor the
Organisation, Unity and Coherence scores for the handwritten
essays.
Spelling Ability
– Not predictive <strong>of</strong> any WOLD score, except Grammar & Usage in
the <strong>spell</strong>-<strong>check</strong> condition (23%, p=0.005)
2. Method
Participants
33 9-11 year-olds (mean age = 10 years, 7 months)
Procedure
– Children wrote three 15-minute essays in response to the WOLD
prompt on 3 separate occasions.
– Tasks: keyboarding with <strong>spell</strong>-<strong>check</strong>, keyboarding without <strong>spell</strong><strong>check</strong>
and handwriting.
– <strong>The</strong>y also completed the 'Alphabet Fluency Task' on the
keyboard(1st visit) and the BAS-II Spelling Test (3rd visit).
Scoring Criteria
WOLD (CQ)
– 6 criteria* (marks out <strong>of</strong> 4) from the standard mark scheme, with
an overall potential score <strong>of</strong> 24. *'Ideas & Development', 'Organisation,
Unity & Coherence', 'Vocabulary', 'Sentence Structure & Variety', 'Grammar &
Usage' and 'Capitalisation & Punctuation‘
– Inter-rater reliability for all categories and overall WOLD was
greater than or equal to 0.8.
Script Errors
– Six error categories (each error was only counted once):
• Spelling: simple, word choice (substitution) & merger
• Other: grammar, total misc, overall total.
– Overall analyses used percentage error in whole text.
BAS-II Spelling Test:
– Raw score <strong>of</strong> number correct
Keyboarding Fluency:
– Letters per minute
Percentage Errors
Transcription Method
– Percentage Spelling, Grammar and Overall Errors:
• No Spell-<strong>check</strong> and Handwritten > Spell-<strong>check</strong>
Typing Fluency
– Predicted 14% <strong>of</strong> total other errors in the <strong>spell</strong>-<strong>check</strong> condition,
indicating that greater keyboarding fluency means fewer errors.
Spelling Ability
– Predicted at least 25% <strong>of</strong> the variance in percentage <strong>of</strong> Spelling
and Overall errors for the No <strong>spell</strong>-<strong>check</strong> and Handwritten
essays.
4. Conclusions
– <strong>The</strong> difference in errors between the handwritten and no <strong>spell</strong><strong>check</strong>
essays indicates that children may ‘sub-contract’ <strong>spell</strong>ing
to the computer when keyboarding (even when no aid is
present).
• Implications for keyboarded exam scripts and different writing
strategies depending on transcription method (handwriting vs.
keyboarding).
– CQ was constrained by children’s relatively low keyboarding
fluency.
– Future research looking at older children, with greater
keyboarding fluency, could help clarify the <strong>effect</strong>s <strong>of</strong> transcription
methods found in this study.
Contact: Kirsty Walter, Psychology Department, Oxford Brookes University
E-mail: kwalter@brookes.ac.uk

References
Connelly, V., Gee, D., and Walsh, E. (2007). A comparison <strong>of</strong> keyboarded and handwritten compositions
and the relationship with transcription speed. British Journal <strong>of</strong> Educational Psychology, 77, 479-492.
MacArthur, C. A., Graham, S., Haynes, J. B., and De La Paz, S. (1996). Spelling <strong>check</strong>ers and students
with Learning Disabilities: Performance comparisons and impact on <strong>spell</strong>ing. <strong>The</strong> Journal <strong>of</strong> Special
Education, 30(1), 35-57.
Torrance, M., and Galbraith, D. (2008). <strong>The</strong> Processing Demands <strong>of</strong> Writing. In C. A. MacArthur, S.
Graham, and J. Fitzgerald, Handbook <strong>of</strong> Writing Research (pp. 67-80). New York: Guilford Press.